Medical records and kids' phone numbers turned up in Google results

Fun fact: over 60% of all workshop injuries are the result of using a table saw without really knowing how to use a table saw. We're talking somewhere between 30,000 and 40,000 cuts, slices and amputations, every year — and that's just in North America. Taking the time to learn how to safely use the saw before firing it up could save you from having to learn how to text without any thumbs. The same goes for any other tool you might use to make your way. Artificial Intelligence, for example. According to Futurism, a whole lotta personal information and intimate chats with Anthropic's Claude AI have been ending up online and are pretty searchable if you know where to look — stuff that folks might not want other folks reading. But Anthropic has stated that this isn't due to a bug. It's a feature that users have to activate in order to use. Hitting Claude's share button creates a "public" link that anyone holding it can view.

The warning, though, doesn't alert users that shared content could wind up being indexed by a search engine — meaning that a user could effectively be sharing their Claude-generated document with the entire internet.

Among the material Futurism turned up: a patient's medical report, clinical trial results listing patient names, phone numbers for primary school-aged children, company files marked internal-use-only, and employee reviews. Artifacts, the feature that saves a project as a standalone app, at least warns you up front that publishing one makes it visible in search results.

Some of that stuff is getting indexed by Google. That's good news if you intended to create a vibe-coded app for folks to enjoy. But not so great if you were talking corporate secrets with Claude or using it to transcribe medical records. If you must use AI or — despite all of the reasons why you shouldn't — want to, take the time to learn about the ramifications of how turning on features in your LLMs can and can't screw with your life.